Proceedings of IMECE2008

2008 ASME International Mechanical Engineering Congress and Exposition

October 31-November 6, 2008, Boston, Massachusetts, USA

IMECE2008-69122
INTEGRATED CONCEPTUAL DESIGN ENVIRONMENT FOR
CENTRIFUGAL COMPRESSORS FLOW PATH DESIGN
Leonid Moroz Yuri Govorusсhenko
SoftInWay, Inc. SoftInWay, Inc.
Petr Pagur Leonid Romanenko
SoftInWay, Inc. SoftInWay, Inc.

ABSTRACT INTRODUCTION
A new method for centrifugal and mixed-type Turbomachinery flow path creation using an
compressors flow paths design based on a unique integrated conceptual design environment (IE) allows the
integrated conceptual design environment is presented in designer to shorten the design development process
this article. At the heart of this new method is the significantly, thereby decreasing engineering costs and
translation of proven, integrated design environments improving productivity. It gives an opportunity to review
that have been successfully used for axial a large number of variants and design parameters to
turbomachinery for many years. realize optimum results. [1-4].
This integrated environment is a seamless and swift This article is devoted to describing new a approach
processing scheme that incorporates stages aerodynamic for developing turbomachinery design systems, IE
analysis and preliminary design/sizing based on the one- components that work for various turbomachinery design
dimensional method interactive spatial blade profiling, platforms, and new subsystem elements for radial
export of blade geometry to CAD and CFD tools, 3D turbomachinery conceptual design.
stress and vibration analysis, and finally, flow modeling.. Initially, IE was developed for axial turbines (mainly
The design process is demonstrated for a centrifugal steam). Later, it was expanded for gas turbines
compressor design utilizing AxSTREAM software. (especially blade cooling calculations) and axial
compressors via plug-in modules. The new challenge
NOMENCLATURE AND GLOSSARY designers face today is developing mixed flow
Variables machinery. These new design system must be flexible
G mass flow rate and have the capabilities to develop axial, radial and
Ht = Δucu/u2 load factor mixed flow machinery at the same IE. As a result of our
R reaction research, two main requirements were specified for the
D diameter system: the majority of modules must be compatible with
P pressure every type of turbomachinery, and specific modules must
T temperature be able to run simultaneously (axial and radial turbine,
PR pressure ratio axial and radial compressor). It turned out that invariant
l blade length modules (project data access, graphical display of
n speed of rotation information, multi-choice calculation and optimization,
u tangential velocity import/export, etc. possibilities) can work for any
α flow angle in absolute frame platform. Also, we determined that conceptual design
eff_tt isentropic(total-to-total) efficiency modules for centrifugal and mixed flow machines would
fit extremely well within the IE structure prototype that
Subscripts worked for axial machinery. Concurrently this IE
0 at the stage inlet, at the beginning of approach and integration can be extended to blade
the process (impeller) 3D profiling, stress analysis, and 3D Flow
1,2,3 in the 1,2,3 sections analysis.
s meridian In the first section of this article, the architecture of a
z axial new integrated system for turbomachinery conceptual
tt total to total design is described. Then, the preliminary design
Superscripts procedures of the radial turbomachinery stages are set
• stagnation parameter forth, followed by 1D calculations on design and on off-
design mode. The next part presents 3D blade design

1 Copyright © 2008 by ASME

procedures, 3D potential flow analysis, and stress and - involve a set of design modules necessary for design
vibration analysis. The concluding section of the article procedures under one operating platform (an umbrella,
brings all components together by presenting examples per se) that performs initial sizing and optimization, 1D
of centrifugal compressor design. formulations, and builds blade 3D geometric models that
are available for final refinement by means of 3D
aerodynamic and stress analysis;
- give the ability to automate multivariate and
optimization calculations using embedded models;
- provide flexibility in carrying out interactive design
INTEGRATED CONCEPTUAL DESIGN scenarios including rollback, versions support, project
ENVIRONMENT integrity, etc
An effective system for turbomachinery flow path - ensure expandability, scalability, and maintainability;
conceptual design needs to do the following:

Flowpath
Preliminary
3D Geometry Design
Export Optimization Inverse Stage
Subsystem Calculation
and
Blades Stress Optimization
and Modal
Analysis

1D Flowpath
3D Project database Analysis
Blade
Design and Optimization

Profile database
Cascade
Profiling
Materials
Fluid models 2D Flowpath
Loss models Analysis
DOE, and Optimization
…
Optimization and
Multivariant
Analysis

3D
Database
Potential Flow
Management Data Manipulation Analysis
Subsystem and Visualization

User Interface

Invariant Axial Turbine component Axial Compressor component Radial turbine component Centrifugal Compressor component

Fig. 1. The architecture of the turbomachinery flowpath integrated conceptual design system

- provides users with convenient mechanisms to input,
edit, display data and export data to other systems.
In Fig.1, the architecture of the turbomachinery flow
paths integrated conceptual design environment is
shown. It can be seen that more than two thirds of the
subsystems are invariant (capable of working various
platforms while maintaining seamless processing). Thus,
after new problem definitions are added into the system
by the designer, the program gains total access to
existing design subsystems.

2 Copyright © 2008 by ASME

 CENTRFUGAL COMPRESSOR ANALYSIS
Traditionally, for centrifugal compressor operation
analysis for design and off-design points, the verification
analysis problem has been performed in 1D
formulations.
In verification analysis of a centrifugal compressor,
flow is treated as one-dimensional only [6]. Obviously,
there is significant 3D flow in a centrifugal compressor.
Considering the importance of quick review of design
results, a 3D potential flow solver was developed and
embedded into the system.
In contrast to 3D viscous flow calculations in a flow
path, simplified 1D methods provide a close
approximation for experimental data depending on the
quality of empirical methods used to determine losses
and deviation angles.
Due to the absence of reliable energy loss and flow
Fig.2. Centrifugal compressor [7] flow path draft
deviation angles in cascades, there is a significant
challenge in developing radial turbomachinery
verification calculation algorithms. In practice, [1] the 3.5
various known construction losses obtained from
experimental data are used.
Centrifugal compressor flow path losses and outlet 3.0

flow deviation angles are estimated based on existing

data in literature [5,6]. Though obtained results can be 2.5
characterized as satisfactory, loss models require further
PR_tt

refinement One way of increasing the reliability of

results is the use of custom libraries for proven loss 2.0

models within the integrated design system environment.

The typical approach to solve this task and boundary 1.5
condition is presented in [6]. The authors of this paper
used the same method, so it is not necessary to explain it.
1.0
As an example, validation of the 1D solver was
0.5 1.0 1.5 2.0 2.5 3.0
performed based on experimental data presented in [7]. Gcor, kg/s
Flow path of the compressor shown in Fig.2 was
studied with both a vaned and a vaneless diffuser. Fig.3. RADIVER compressor performance map with
Performance calculations show a close correlation of the vaneless diffuser. (-o- experiment, -.- calculations).
obtained efficiency with experimental data for the at 80% rotation speed.
vaneless diffuser design in a quantitative sense and a
0.95
close correlation for vaned diffuser design in a
qualitative sense (Fig.3). Efficiency levels at 80%
rotation speed coincide closely with experimental data as
Impeller Eff_tt

well. (Fig.4, Fig.6). The literature only cites CFD 0.90

calculations results (Figs.7a,7b,7c,7d) and experimental

data at an 80% rotation speed ([9], [10] etc.). These
results correlate closely to experimental data at an 80% 0.85
speed as well.

0.80
1.7 1.8 1.9 2.0 2.1 2.2 2.3
Gcor, kg/s

Fig.4. RADIVER compressor total-to-total

efficiency map with vaneless diffuser (-o- experiment, -.-
calculations) at 80% rotation speed.

3 Copyright © 2008 by ASME

 4.0

3.5

3.0
PR_tt

2.5

2.0

1.5

1.0
0.5 1.0 1.5 2.0 2.5 3.0
Gcor, kg/s

Fig.5. Radiver performance map with vaned diffuser

(-o- experiment, -.- calculations).
Fig. 7a. Static and absolute total pressure comparison
By means of numbers, rotor rotation speeds are at plane 2m at 80% rotation speed. Experimental data
shown in the figure. [10], CFD calculations data [7]

Fig.6. RADIVER compressor total-to-total efficiency Fig. 7b. Isentropic efficiency comparison at plane 2m
map with vaned diffuser (-o- experiment, -.- calculations) at 80% rotation speed. Experimental data [10], CFD
at 80% rotation calculations data [7]

4 Copyright © 2008 by ASME


Fig. 7c. Total pressure ratio (πt) and isentropic
efficiency (η) comparison (r4/r2=1.14) Experimental
data [10], CFD calculations data [8]
Fig. 7d. Total pressure ratio (πt) and isentropic
efficiency (η) comparison (r4/r2=1.14) Experimental
data [10], CFD calculations data [8]
CENTRFUGAL COMPRESSOR SIZING
There exists an opinion in the literature that
preliminary design of turbomachinery stages may be
performed based on experimentally generalized
diagrams. For axial turbines, Smith’s diagram is often
used [9], for radial – Chen and Baines [10].
These diagrams, when compared with solutions
generated according to the proposed algorithms of the
preliminary design, show that utilization of diagrams will
not produce an optimal construction because:
- generalized parameter ranges are not sufficiently
large;
- real, current day design solutions may fit outside the
accepted ranges of these diagrams that were created
several decades ago;
- there is a design results divergence (about 10% at
efficiency) at the same generalized parameters.
The new IE approach gives the designer a uniform
tool for exercising an optimal parameters search of the
axial and radial turbomachines for both turbines and
compressors. The essence of the method is in the
multiple (sequential) solving of the flow path inverse 1D
gas-dynamic calculation problem to find the best
solution. See more details [4]. One variant design is
performed by assigning proper set of flow paths, with
both dimensional (diameter, height) and dimensionless
(flow factor) characteristics. An optimal solution search
is carried out using quasi random search.
In this paper we are introducing an approach for
centrifugal compressors for the first time. For centrifugal
compressors, dimensions determination can be assigned:
inlet parameters P1*, T1*, mass flow rate G, pressure
ratio P3/P1. It is important to know the diffuser type
(vane, vaneless, etc.), Speed of rotation n, rotor diameter
at the stage inlet D1, mass flow rate coefficient c1z/u1,
diameters ratio D2/D1, and flow outlet angle at the stage
α3 are variable parameters. Optimal design criteria can be
maximum stage efficiency or maximum output (in
turbine) or minimum input (in compressor) capacity. If
there exists a reliable losses model procedure, the inverse
1D calculation allows for fast generation of solutions
with sufficient admission of variables, and the ability to
compare design variants to choose the best solution.
Efficiency – flow rate coefficients and typical load
diagrams are shown in Fig 8. Point color corresponds to
efficiency level. Pale points are outside the range of
established constraints for an acceptable solution search
such as the maximum outer diameter constraint. The
envelope of points, including all “candidates” for an
optimal solution, is smooth, but one can see that points
delaminate essentially “into the depth.” Particularly Fig.
8.b shows that there is an efficiency optimum by load
coefficient in the admitted range and efficiency; flow
rate coefficient relationship is neutral in the assigned
range.
5 Copyright © 2008 by ASME
 3D BLADES DESIGN

Approaches to blade profiles are based on a definition of

hub/shroud meridional trajectories with quadratic Bezier-
curves with linear extensions. Intermediate stations
(trajectories) are created as a channel/sub-channel
equidistant curve. (Fig. 9)

Figure 8.a

Fig.9. Meridional projection

The frontal camber map is built as an m’-theta plane.

The angles are preserved on the leading and trailing
edges while intermediate points are distributed along the
quadratic Bezier-curve (Fig.10).

Figure 8.b
Fig.8. Estimated diagrams for centrifugal compressor
efficiency (phi – flow coefficient, psi – work
coefficient)
Source: AxSTREAM.

Fig.10. Frontal m'-theta plane

Spatial shape of blade camber surface is defined with

conformal mapping of meridional trajectories on the

6 Copyright © 2008 by ASME

 m'-theta plane. Camber line spatial point P is defined
with coordinates (x, R, Theta) where x and R are its
coordinates appropriative on meridional view. The theta
coordinate is defined as the intersection of the camber
map on the m'-theta plane and the circular arc with the
radius equal the distance to point on meridional
projection along the curve.

Operational/manufacturing issues are also considered

in a blade design. The availability of radial elements is
critically important for high-speed wheels.
Stacking of 3D sections is controlled with the angular
(theta) offset of LE points of hub, shroud, and any
intermediate sections.
Additionally, TE points of shroud, and any
intermediate sections also may have an angular (theta)
offset.
Blade design procedure allows the use of rounded
(circular) edges and truncated ones (cut-off design). Fig.12. Wheel with independent splitters
A centrifugal compressor design example is shown in
Fig.11.
The embedded capability to export system cascade 3D STRESS ESTIMATION
geometry into popular CAD formats and CFD packages, Aerodynamics and strength requirements lay down
permitting easy traversing all the way from preliminary additional criteria to meet design objectives. That is why
design to 3D gas-dynamic and stress analysis of the strength and stress estimations should be performed
wheel . during the early stages of the design process.
Designing a wheel with splitters often improves the Historically, detailed stress calculations have been very
performance of centrifugal compressors. New design time consuming, slowing down the process of obtaining
tools allows the use of “dependent” splitters (the splitters an optimal project solution. New design tools provide
and blades are located with even angular pitch) and real time stress analysis in seconds.
“independent” splitters when splitter’s leading edge is
located in arbitrary angular offset (Fig.12).
The splitter camber curvature and LE/TE position of
independent splitter may be changed with click-and-drag
of control points as may also be done for main blade.

Fig. 13. Stress (von Mises) in the centrifugal

compressor blade.
Fig.11. Designed centrifugal compressor blade

7 Copyright © 2008 by ASME

 Use of beam theory for radial machine blades is 9000Hz correspond to blade forms of natural vibrations.
impossible because blade deformations do not conform Horizontal lines f1 – f4 show first 4 frequencies of a
to the hypothesis of linear distribution of displacements single blade.
on a cross-section. In spite of the fact that shell theory is
more comprehensive to a blade’s deformation character,
it does not allow one to perform specified calculations
and does not agree with the three-dimensional body.
That is why FEM was chosen for stress and vibration
analysis. There are more details regarding this approach
in “The Finite Element Method” [11]. With a minimum
number of quadratic brick finite elements, the results are
similar to those from calculations obtained with rods and
shells theories. If the number of finite elements are
increased, higher levels of calculation accuracy will be
obtained.
Early in the design process, the aim of stress analysis
is not to determine the true stress field but to identify the
best project variant among the possible solutions that Fig. 15. The first 50 natural frequencies of the
match the aerodynamic criteria. As the final project nears compressor bladed disk
completion, accuracy must be increased. To ensure that
hub and rim are included in the calculations (taking into
account rounded radiuses), the performed calculations
can be made using one symmetric airfoil or the whole
rotor.

Fig. 14. Stress (von Mises) in the compressor’s

blades and hub.

Single blade (Fig. 13) and bladed disk (Fig.14) von

Mises stress calculation results comparison shows that
stresses calculated for single blade differ from blade in
bladed disk up to 30%. In comparison, the first
calculation requires a fraction of a second as compared
to the second one that takes a few minutes in normal
formats .
In Fig.15 the first 50 natural frequencies of bladed
disk are shown. Frequencies thickening near 5000 and

8 Copyright © 2008 by ASME

 3D POTENTIAL FLOW ANALYSIS
3D potential flow analysis [12] is performed to
analyze gas field velocities in the channels to determine
aerodynamic forces on compressor blades. It is assumed
that the flow is potential in a fixed coordinate system.
Viscosity absence, external forces potentiality, and
channel inlet gas vorticity absence are sufficient
conditions to determine flow potentiality.
A flow potential is defined via FEA. The pressure
field can be defined from the Cauchy-Lagrange integral.
Gas compressibility is taken into account iteratively by
the way of flow potential determination with non-
uniform density. The iterative process is convergent for
subsonic flows.
Fig. 16 shows the relative velocities field of the
centrifugal compressor channel, calculated according to
the gas potential flow model. The model allows one to
determine flow degree of irregularity in the channel.
In Fig. 17 static pressure field calculated according to
the same model is shown. This field is used to estimate
bladed disk strength. Fig. 17. Field of the static pressure in the compressor
channel.

CENTRIFUGAL COMPRESSOR
DESIGN EXAMPLES
Parameters for CC with vane diffuser design are:

P1* = 60 KPa,
T1* = 300 K,
n = 25000…35000 rpm
G = 1.5 kg/s
P3/P1 = 3.5
D1 = 90…110 mm
c1z/u = 1...1.5
D2/D1 = 2.0...3.0
α3 = 80o...100o
D4/D3=1.3

Optimal stage compressor flowpath with vaned

diffuser draft is shown on the Fig.18a, velocity triangles
Fig. 16. Field of the relative velocity of the are on the Fig.18b.
compressor channel. The following parameters are received in an optimal
point:
n = 29000 rpm,
c1z/u1 = 1.01,
Ht = 0.86,
B2 = 95.8.
Eff=91.2%

9 Copyright © 2008 by ASME

 a
a

b b
Fig.18. Velocity triangles of the centrifugal Fig.19. Velocity triangles of the centrifugal
compressor rotor with vaned diffuser. compressor rotor with vaneless diffuser.

CONCLUSIONS
Parameters for CC with vane diffuser design are the This article introduces a new approach to the
same. development of turbomachinery design systems. The
Optimal stage compressor flow path with vaneless approach gives the designer the opportunity to design
diffuser draft is shown in the Fig.19a, velocity triangles axial, radial and mixed flow turbomachinery using the
are on the Fig.19b. same tool.
The following parameters are received in an optimal Features of radial turbomachinery integration into the
point: turbomachinery conceptual design software environment
n = 35000 rpm, have been examined in this article. The benefits of this
c1z/u1 = 1.03, approach are unification of the theory, methods, design
Ht = 0.86, procedures, and total functionality for different types of
B2 = 83.0
machines with fast calculations to improve human
Eff=85.7%
productivity.
Results presented in the article are obtained using Through a single interface with embedded (plug-in)
turbomachinery flow path conceptual design suite modules, the design process successfully traverses the
AxSTREAM [2]. way from preliminary design estimates through detailed
optimization via multivariate calculations. The final
phase culminates with the development of three-
dimensional blades models with FEA stress and CFD
flow analysis at minimum time.
Work in the IE also gives other useful opportunities
such as optimization and takes into account weight and
strength criteria, in addition to mixed type flow path
analysis.

10 Copyright © 2008 by ASME

 The IE incorporates an algorithm of the radial The following copyright and trademark-protected software were used in
this paper:
turbomachinery rotor, stress, and modal analysis that can
- AxSTREAM 2.3 Copyright 2002-2008 SoftInWay Inc.
be easily adapted to meet the needs of preliminary design - AxPLAN 1.0 Copyright 2002-2008 SoftInWay Inc.
in terms of accuracy and speed of results.
These new theories and tools will substantially
increase the efficiency and reliability of radical and
mixed flow turbomachinery.

REFERENCES
1. Biba Yu., Menegay P. “Inverse Design of Centrifugal
Compressor Stages Using a Meanline Approach”.
International Journal of Rotating Machinery, 10: 75–84,
2004
2. Moroz. L., Govorushсhenko Y., Romanenko L., Pagur
P.. “Methods and Tools for Multidisciplinary
Optimization of Axial Turbine Stages with Relatively
Long Blades”. Proceedings of Turbo Expo 2004. 14-17
June Vienna, Austria.
3.Baines N. “Radial Turbines: An Integrated Design
Approach”. Concepts NREC, 2005.
4. Moroz. L., Govorushсhenko Y., Pagur P. “A uniform
Approach to Conceptual Design of Axial
Turbine/Compressor Flow Path”. The Future of Gas
Turbine Technology 3rd International Conference 11-12
October 2006, Brussels, Belgium.
5. Eckert B. ”Axialkompressoren und Radialkompres-
soren. Anwendung/Theorie/Berechnung”. – Springer-
Verlag, 1953.
6. Aungier R.H. “Centrifugal compressors: a strategy for
aerodynamic design and analysis”. - New York, 2000.-
315 pp.
7. Carsten Weiß, Daniel R. Grates, Hans Thermann,
Reinhard Niehuis. Numerical Investigation of the
Influence of the Tip Clearance on Wake Formation
Inside A Radial Impeller. GT2003-38279. Proceedings
of ASME Turbo Expo 2003: Power for Land, Sea, and
Air. June 16–19, 2003, Atlanta, Georgia, USA
8. Pavel E. Smirnov, Thorsten Hansen, Florian R.
Menter. Numerical Simulation of Turbulent Flows in
Centrifugal Compressor Stages with Different Radial
Gaps. GT2007-27376. Proceedings of GT2007ASME
Turbo Expo 2007: Power for Land, Sea and Air. May
14-17, 2007, Montreal, Canada.
9. Smith S.F. “A simple correlation of turbine
efficiency”. Journ R Aeronaut Soc 69: 467-470.
10. Chen H, Baines N. “The aerodynamic loading of
radial and mixed flow turbines”. Int Journ Mech Sci 36:
63-79, 1994.
11. Zienkiewicz O.C, and Taylor R.L. “The Finite
Element Method”. 4-th ed., McGraw-Hill, London, UK,
1991.
12. Fletcher C.A.J. “Computational Techniques for Fluid
Dynamics 2”. - Springer-Ferlag,1988.

11 Copyright © 2008 by ASME